APC Smart-UPS info
The following information was provided by der Mouse
<mouse@Rodents.Montreal.QC.CA>
and appears here with his
permission.
The info about the S shutdown is of special interest to
me...
if I can verify its behaviour during an extended shutdown I'll
modify my apcmon script to use it.
--sjg
Date: Wed, 2 Apr 1997 15:39:17 -0500 (EST)
From: der Mouse
Message-Id: <199704022039.PAA26760@Twig.Rodents.Montreal.QC.CA>
To: sjg@crufty.net
Cc: mouse@Twig.Rodents.Montreal.QC.CA
Subject: Re: UPS unattended shutdown
> See http://www.crufty.net/sjg/apcmon.html
I arranged access to a Smart-UPS 2200XL and I find that:
- You don't need APC's cable. The book that comes with the UPS
documents two of the lines as having RS-232 signal levels; it turns
out that a cable that hooks them to pins 2 and 3, and ground to
ground, works Just Fine for me. (DB9M pin 1 to DB25M pin 2; DB9M
pin 2 to DB25M pin 3; DB9M pin 4 to DB25M pin 7.) The 2200XL is
supposed to be a "smart signaling" UPS, so I assume the black cable
(plus an appropriate DB9-DB25 converter) would have this or a
similar pinout.
- The interface, at least for the 2200XL, is a lot more complicated
than the one your scripts imply. I don't know it all yet, but I
know enough to be useful. :-)
- Yes, I read the note in the front of the PowerChute book (another lab
whose sysadmin I'm on good terms with uses the real thing); however,
I carefully didn't use their package, so I didn't thereby agree not
to reverse-engineer anything.
I'm sending this to you 'cause you're the best (indeed, only)
repository of UPS info I know of. :-)
Communication is at 2400 baud.
The UPS is in one of three states at any time:
- Dead state: no lights, no response at all to serial line.
- Dumb state: RS232 receive line is shutdown signal (high for 4.5
seconds initiates shutdown); RS232 send line is powerfail signal.
Additionally, the RS232 line is monitored for a 0x59 (Y) character,
which is the signal to switch into smart mode. As far as I can
tell nothing but Y or the 4.5-second signal gets the UPS's
attention in dumb mode.
- Smart state: RS232 send and receive lines are normal RS232. I don't
know whether a 4.5 second break condition will initiate a shutdown.
Transitions as following are possible; as far as I know no others are
possible:
Dead->dumb: front-panel-initiated power-up, or in some cases mains
power returning
Dumb->smart: receiving the Y command
Dumb->dead: front-panel-initiated power-down, also apparently some
timeout (completely drained batteries?) when in
4.5-second-break-initiated shutdown mode.
Smart->dumb: receiving the R command; also, front-panel-initiated
power-up from a software-initiated shutdown state seems to do this.
Smart->dead: Some timeout when in a shutdown state, even if batteries
aren't completely drained.
The rest of this document describes only smart mode.
Normally, the UPS is silent except for replies to commands. However,
there are a few characters it will gratuitously send. Here are my best
guesses at them:
! Sent periodically (approx. every 30 seconds) while running on
battery.
$ Sent, once, when returning from on-battery to mains power.
? Sent, once, on front-panel-initiated power-up (and enters dumb
mode).
| I've seen it but I don't know what it means. It may have
something to do with worn-out batteries.
%, + Unsure. Low battery seems to send alternating % and + every
few seconds while beeping like mad. I have seen % and + in
other contexts though.
In the following list of commands, "adjustable" means the + and -
commands do something useful, usually cycling through a small number of
values; "prog-adjust" means "adjustable, but only in PROG mode" (see
the 1 command). Prog-adjust values usually adjust a two-nibble hex
value.
The first column is the character, either the character itself for
space-through-tilde or a two-digit hex code prefixed with 0x for
characters outside that range. I have tried all 256 single-byte
commands; all byte values not listed here consistently reply NA.
"reply" means that the indicated string is sent, followed by a CR/LF
pair. Except for values noted as adjustable, prog-adjust, or whose
description explicitly mentions otherwise, the + and - commands return
NO.) When multiple "reply" lines are given, this means I have seen
each of those strings under at least some circumstances. Some commands
have [...] comments after the ->; these indicate semantics that some
source has attributed to that value and I have seen no reason to
disbelieve. "S shutdown" refers to the power-down state after a
shutdown initiated by the S command. In this state, six of the
front-panel lights do a cute marquee-like pattern.
0x01 -> reply SMART-UPS 2200 XL
(Presumably this is some sort of self-identify command.)
0x0e -> no reply
0x1a -> reply #uD43132135138129uM43229234239224uA43110112114108uI43253264271280lD43103100097106lM43177172168182lA43090092081085lI43196188208204e44200155090oI43230240220225oD13115oJ13100oM13208s441HMLLq44202050710p443020180300600k4410TLNr443000060180300E443336168ON OFF
(I have no clue what this string means.)
' -> reply 00
+ -> if last value is adjustable, adjust upwards and reply with
two-nibble hex (PROG mode) or OK (non-PROG mode). Otherwise,
reply NO. (In non-PROG mode, only one + or - is allowed before
re-issuing the display command.)
- -> if last value is adjustable, adjust downwards and reply with
two-nibble hex (PROG mode) or OK (non-PROG mode). Otherwise,
reply NO. (In non-PROG mode, only one + or - is allowed before
re-issuing the display command.)
0 -> reply 9F (prog-adjust)
1 -> no reply; it eats anywhere from zero upwards of the following
characters. In some cases (I have no reliable way of provoking
this, but usually can manage it within a few tries) it will
reply PROG, which indicates that the UPS has entered PROG mode.
In PROG mode, + and - can adjust prog-adjust values and can be
given multiple times in succession.
4 -> reply 28 (prog-adjust)
5 -> reply F1 (prog-adjust)
6 -> reply F9 (prog-adjust)
7 -> reply 00
8 -> reply 00
reply 10 (after S shutdown)
9 -> reply FF
reply 00
> -> [battery count] reply 000 (8-bit unsigned, in decimal(!), adjustable)
? -> reply 33AB
reply 33AC
reply 0000
reply 33AD
@ -> no reply, eats next character (?)
reply NA (after S shutdown)
A -> [test LEDs] burn all LEDs, beep, send OK immediately
B -> [battery voltage, volts] reply 55.65
reply 51.85
reply 50.25
(prog-adjust)
C -> [battery temp, deg C] reply 029.2
reply 023.4
D -> reply NO
reply NA (after S shutdown)
E -> reply 336
F -> [freq, hz] reply 60.0
reply 60.00
G -> reply S
K -> no reply, but if repeated "soon" (for some suitable value of
"soon"), replies OK and powers down hard (not to S-shutdown
mode) after p time.
L -> [some voltage?] reply 114.4
reply 000.0
(prog-adjust)
M -> [some voltage?] reply 115.7
reply 000.0
N -> [some voltage?] reply 113.7
reply 000.0
O -> [some voltage?] reply 114.4
reply 119.6
reply 000.0 (after S shutdown)
P -> [load, %] reply 003.6
reply 005.2
(prog-adjust)
Q -> [status, first digit on-battery-p] reply 08
reply 10
reply 00 (after S shutdown)
R -> reply BYE and enter dumb state
S -> reply NA (when running on mains power)
reply OK, pause, shut down and enter S shutdown mode
reply NA (after S shutdown)
U -> [simulate powerfail] reply OK, do power-fail (and power-restore) sequence
reply NA (after S shutdown)
V -> reply MWD
W -> [do self-test] reply OK, initiate self-test
reply NA (after S shutdown)
X -> [self-test status?] reply OK
reply NG (after S shutdown)
reply NO (??)
Y -> reply SM (this is the only command that works when in dumb mode)
if in dumb mode and on battery, the break condition makes the reply
look different - I see 0x00 0xea 0x4d instead of 0x53 0x4d - but
smart mode is entered and a second Y will reply with a clear SM.
Z -> no reply
reply NA (after S shutdown)
a -> reply 3.!$%+?=#|.^A^N^Z+-789>@ABCDEFGKLMNOPQRSUVWXYZ'abcefgjklmnopqrsuxyz~^?
(Note: the ^A, ^N, ^Z, and ^? digraphs above correspond to
0x01, 0x0e, 0x1a, and 0x7f bytes in the actual response.) I
don't know what, if anything, the 3.!$%+?=#|. part means, but
the rest of it appears to be a list of commands that do
something. The only commands that I know do anything that are
not listed there are 0 1 4 5 6 ? (qv); there are no commands
listed there that I do not list here.
b -> reply 81.9.D
(prog-adjust)
(I don't have a clue what this is.)
c -> reply UPS_IDEN
(+ or - makes it eat next 8 bytes and store them here)
e -> [wakeup capacity, %] reply 00
(adjustable with + and -, cycles through 00, 15, 50, 90)
f -> [battery capacity, %] reply 089.0
reply 086.0
g -> reply 048
(I have no clue what this value is.)
j -> reply 0307:
reply 0220:
reply 0185:
reply 0176:
reply 0129:
reply 0118:
reply 0123:
reply 0126:
reply 0116:
reply 0087:
reply 0088:
reply 0105:
reply 0110:
is this remaining on-battery run time at current load, perhaps?
k -> [audible alarm] reply 0
(adjustable with + and -, cycles through 0 T L N)
l -> [low transfer point] reply 103
(adjustable with + and -, cycles through 103 100 097 106)
m -> reply 10/07/96
(PROG-mode + or - makes it eat next 8 bytes and store them here)
n -> reply ES9633162488^@^@
(note the ^@ digraphs correspond to 0x00 bytes in the response.)
(PROG-mode + or - makes it eat next 14 bytes and store them here)
o -> [output volatge on battery] reply 115
p -> [shutdown time, sec] reply 020
(adjustable with + and -, cycles through 020 180 300 600)
q -> [low-battery warning time, minutes] reply 02
(adjustable with + and -, cycles through 02 05 07 10)
r -> [wakeup delay, sec] reply 000
(adjustable with + and -, cycles through 000 060 180 300)
s -> [line sensitivity, H/M/L] reply H
(adjustable with + and -, cycles through H M L L)
u -> [high transfer point] reply 132
(adjustable with + and -, cycles through 132 135 138 129)
x -> reply 10/07/96
(+ or - makes it eat next 8 bytes and store them here)
y -> reply (C) APCC
z -> reply CLEAR, reset to defaults; in PROG mode, resets more values.
~ -> reply 00
(I don't know what this is.)
0x7f -> reply NO
reply OK and power up (after S shutdown)
der Mouse
mouse@rodents.montreal.qc.ca
7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B